Homelessness is a major problem across North America: more than 5 million Americans have been homeless in the last 5 years and 15% of the homeless are parents with children. Homeless people suffer from high rates of illness and frequently encounter barriers to accessing care. Comprehensive information on health care utilization in homeless people would help guide the development of health policy, but these data cannot be obtained through administrative databases in the U.S., because only one-third of the homeless have Medicaid and more than half are uninsured. The aims of this study are: (1) to measure all health care utilization (hospitalizations, emergency department visits, and ambulatory care visits) in a representative sample of homeless mothers, single women, and single men, and to identify individual characteristics associated with health care utilization; and (2) to test the hypothesis that the use of ambulatory care services by homeless people reduces the likelihood of emergency department utilization. This study will be conducted in Toronto, Ontario, where there is both a large homeless population and comprehensive administrative databases that capture the delivery of all health services for every resident. Because homeless people in Toronto have clinical characteristics similar to that of homeless people in the U.S., the results of this study will be relevant to the American setting. This study will enroll a sample of 1,200 homeless people at shelters and meal programs. A baseline survey will obtain data on demographics, health conditions, and barriers to care. Data on health care utilization during the following year will be obtained through linkage to administrative databases. Characteristics that predict frequency of health care utilization will be identified using regression analyses. Factors associated with the likelihood of emergency department visits will be determined using logistic regression models. Age- and sex-specific rates of health care utilization in homeless individuals will be compared to that of the general population. This study will yield data of unprecedented quality that will be useful in developing effective strategies to deliver health care for homeless individuals and families. Findings will be relevant to health care providers, insurers, and others who share a responsibility for improving the health of Americans who are homeless.